Telerehabilitation in the Home Versus Therapy In-Clinic for Patients With Stroke: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Telerehabilitation in the Home Versus Therapy In-Clinic for Patients With Stroke (NCT02360488), a randomized phase 1 study. The primary outcome is Change in Arm Motor Fugl-Meyer Score From Baseline to 30 Days Post-therapy.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group trial with single masking, described in the official title as assessor-blind. The registry identifies the study as completed and lists phase 1. It records the study start as 2015-09 and primary completion as 2018-04. The sponsor was the University of California, Irvine.

The planned enrollment was 124, with the brief summary noting that the number might be larger depending on participant dropout. The reported outcome groups each have 62 participants. The registry describes the intended participants as people with stroke-related arm motor deficits 4-36 weeks after ischemia or intracerebral hemorrhage.

Both groups were assigned 6 weeks of intensive arm motor therapy. The planned schedules were described as identical in intensity, duration, and frequency: 36 sessions, consisting of 18 supervised and 18 unsupervised sessions, with each session lasting 80 minutes including a 10 minute break. The delivery setting differed. Telerehabilitation was delivered at home using an internet-connected computer, while the in-clinic group received supervised outpatient therapy at a study site for half of the sessions and home-based unsupervised therapy for the remainder.

The primary endpoint was within-participant change in arm motor Fugl-Meyer score from baseline to the 30 Day Follow-Up Visit. The posted primary outcome is continuous and reports a mean and standard deviation for each group. The registry also posts a linear regression analysis of the mean difference, identified as a non-inferiority analysis. Details of random sequence generation, allocation concealment, the assessor-masking procedures, and the original sample-size justification are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Intention-to-Treat with multiple imputation of missing outcomes

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: All subjects who were randomized.

6. Primary Endpoint

The registry lists the primary outcome as Change in Arm Motor Fugl-Meyer Score From Baseline to 30 Days Post-therapy, measured over from the Baseline Visit to the 30 Day Follow Up Visit.

The primary endpoint is change in arm motor Fugl-Meyer score from the baseline visit to the 30 Day Follow-Up Visit. The registry describes the scale as a measure of arm motor impairment in the body structure/function domain. It consists of 33 assessments summed into a total score ranging from 0-66; higher scores indicate better arm motor status, with 66 described as normal. No subscores were evaluated.

Change from baseline is used to express how the measured score differs at follow-up from each participant’s starting measurement. Comparing average changes between groups can therefore address whether the two treatment settings were associated with different average changes over the stated follow-up interval. The record identifies this as a non-inferiority trial, so the intended question is not merely whether the observed group means match exactly. A non-inferiority interpretation depends on a prespecified margin defining how much difference would still be compatible with the study’s objective. That margin is not reported in the registry record.

The registry reports group means and standard deviations for the endpoint and identifies the analysis population as intention-to-treat with multiple imputation of missing outcomes. The outcome description does not give the actual imputation procedure or explain how the reported group summaries relate to that analysis population.

7. Results Posted in the Registry

• In-Clinic Therapy: mean 8.36 versus 7.86 on control; difference 0.50 (95% CI -1.92 to 2.92), standardized effect d = 0.07; z-test p = 0.685.

8. Statistical Interpretation

9. Statistical Methods Explained

The endpoint is continuous: each participant’s arm motor Fugl-Meyer score is measured on a numeric scale, and the primary outcome is change between baseline and follow-up. The registry reports the group mean and standard deviation. A mean describes the average observed change in a group; a standard deviation describes how dispersed individual scores are around that average. General statistical explanation: the mean is useful for summarizing the group’s central tendency, while the standard deviation conveys variation among participants. Neither statistic alone describes the full shape of the outcome distribution.

The registry’s own analysis is described as linear regression and reports a mean difference (net). Regression estimates a group contrast while using a model for the outcome. The record does not report which covariates, if any, were included, how the model was specified, or what assumptions were checked. Those details are not reported in the registry record. The analysis is labeled non-inferiority, but the non-inferiority margin is also not reported in the registry record.

The posted recalculation is a simple, unadjusted comparison of the group summaries. It reports a mean difference of 0.5, a 95% confidence interval from -1.916 to 2.916, and an unadjusted p-value of 0.685. The recalculation uses the group means and standard deviations and does not reproduce the registry’s regression analysis. It should therefore be read as a supplemental descriptive comparison, not as a replacement for the sponsor’s analysis.

What does a mean difference describe?

General statistical explanation: a mean difference expresses the contrast between average outcomes in groups. Its sign depends on the order in which groups are compared. For this outcome, the scale’s higher values indicate better arm motor status, but the direction and meaning of a difference still depend on the contrast definition and on whether the outcome is change from baseline. The posted recalculation compares the in-clinic mean with the telerehabilitation mean. The registry’s reported regression estimate is separately described as a net mean difference, and details of its coding and adjustment are not reported in the registry record.

What does a p-value address?

General statistical explanation: a p-value describes how compatible the observed data are with a specified null hypothesis under the statistical model. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. The registry reports p = .96 for its regression analysis, while the unadjusted recalculation reports p = 0.685. These values belong to different analyses. Neither is, by itself, the non-inferiority decision, which requires a margin and a specified decision rule.

Why is non-inferiority different from a test of no difference?

General statistical explanation: a conventional test of no difference asks whether the data are inconsistent with an exact zero difference. A non-inferiority test asks whether the data exclude an unacceptable loss relative to a prespecified margin. The conclusions can differ because they use different hypotheses and decision criteria. For this trial, the registry identifies the sponsor’s analysis as non-inferiority, but does not report the margin. The result therefore cannot be fully evaluated against that criterion from the registry record alone.

What does multiple imputation mean?

The registry describes the primary analysis population as intention-to-treat with multiple imputation of missing outcomes. General statistical explanation: multiple imputation uses a model to represent uncertainty about missing values rather than treating a single filled-in value as known with certainty. The validity of the approach depends on assumptions about the missing data and on the imputation model. The registry record does not report the amount or pattern of missing outcome data, the variables used in imputation, the number of imputed datasets, or the assumptions used. It is therefore not possible to assess those features from the posted record.

The registry also reports group-level outcome counts of 62 in each arm. Those counts do not, by themselves, establish that all randomized participants had observed follow-up outcomes, because the primary population is described as including imputation of missing outcomes. The relationship between group counts, missing observations, and the imputed analysis is not reported in the registry record.

Why does over-dispersion matter?

General statistical explanation: over-dispersion is a concern when observed variability is greater than a statistical model expects. It can make standard errors too small if the model does not account for the extra variability, which can in turn make confidence intervals too narrow and p-values too small. This concept is especially relevant to models for counts or rates; the primary endpoint is continuous, and the registry reports standard deviations. Whether any model diagnostics or adjustments for variance were used is not reported in the registry record. The term should not be taken as evidence that over-dispersion was present in this trial.

More broadly, statistical results depend on the definition of the outcome, the analysis population, the handling of missing observations, and the model. The registry provides the endpoint definition, group summaries, the broad regression method, and the population description, but not enough detail to reconstruct the sponsor’s regression or imputation analysis. The distinction between reported analysis and unadjusted recalculation is therefore important when interpreting the numbers.

10. Confidence Intervals and Precision

The registry’s regression estimate is a mean difference (net) of 0.06, with a 95% confidence interval from -2.14 to 2.26. The interval crosses zero and extends in both directions, so the estimate is compatible with a range of possible group differences rather than a single precisely established value. The interval also cannot be judged against a non-inferiority margin because that margin is not reported in the registry record.

The unadjusted recalculation gives a mean difference of 0.5, with a 95% confidence interval from -1.916 to 2.916. This interval also crosses zero and includes possible differences in either direction. It is based on the posted group summaries and is distinct from the sponsor’s regression analysis. The two intervals should not be combined or treated as estimates from the same method.

General statistical explanation: a confidence interval describes a range of values compatible with the data and analysis procedure at the stated confidence level. In repeated use of the same procedure under its assumptions, intervals constructed this way would cover the target value at the stated rate. A confidence interval is not a statement that there is a 95% probability that this particular interval contains the true value.

The observed standard deviations, 6.68 for telerehabilitation and 7.04 for in-clinic therapy, describe participant-level spread in the reported change scores. The interval widths reflect both variability and the amount of information in the analysis, as well as the analysis method. The registry does not provide the sponsor model’s full specification, so the contribution of adjustment or other model choices to its interval cannot be determined. The intervals are informative about uncertainty, but they do not by themselves resolve the non-inferiority question.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed effect, not the trial’s original design assumptions. For this continuous-outcome comparison, the figures are participants per arm, not total events. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect.

The 80% power column gives the number of participants per arm that the planning calculation estimates would be needed to have 80% power under those assumptions and for the specified effect. The 90% power column gives the corresponding number per arm for 90% power. Power is the probability, under the assumed effect and model, that a study would meet its specified statistical criterion. These are planning quantities, not descriptions of what happened in the completed trial.

The table includes scenarios planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. The corresponding figures are 2,957 and 3,959 participants per arm for the 100% scenario; 5,257 and 7,038 participants per arm for the 75% scenario; and 11,828 and 15,835 participants per arm for the 50% scenario, for 80% and 90% power respectively. Smaller effects require more participants under the same planning assumptions because they are harder to distinguish from random variation.

This is not observed power. It is a forward-looking illustration calculated from an observed effect and stated assumptions. It does not show the trial’s actual power, does not validate the original study design, and should not be interpreted as the original enrollment target. The original sample-size justification and its assumptions are not reported in the registry record. The plotted curve should be read as a visual representation of how the planning requirement changes with the effect assumed, not as evidence about outcomes beyond the listed scenarios.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.500; pooled standard deviation σ = 6.862.
2. n = 2 × (1.96 + 0.842)² × 6.862² / 0.500² = 2,957.0, rounded up to 2,957 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.499999999999999                      # observed difference in means
sigma <- 6.86236110970561                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2360488)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 2,957 (80% power), 3,959 (90% power)
 75% of observed effect: n per arm = 5,257 (80% power), 7,038 (90% power)
 50% of observed effect: n per arm = 11,828 (80% power), 15,835 (90% power)
power.t.test: 2957.9 
Simulated power at n = 2,957 per arm: 0.794 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The sponsor’s own posted result is a linear regression analysis identified as non-inferiority. It reports a mean difference (net) of 0.06, a 95% confidence interval from -2.14 to 2.26, and p = .96. This is the primary reported analysis to use when describing the registry’s result.

The unadjusted recalculation from the group summaries reports a mean difference of 0.5, a 95% confidence interval from -1.916 to 2.916, and an unadjusted p-value of 0.685. It is a simple comparison based on the posted means and standard deviations, not a reproduction of the regression model. The estimates and intervals differ because the analyses are not the same. The registry does not provide enough detail about model specification or adjustment to reconstruct the sponsor’s estimate.

The discrepancy indicator is false. Even so, the sponsor’s regression result and the recalculation should be kept distinct: the former is the analysis reported by the registry, and the latter is an unadjusted supplementary calculation. The non-inferiority margin is not reported in the registry record, so neither result can be assessed against that margin using the posted information alone.

14. Multiplicity, Missing Data and Analysis Populations

The registry reports one primary outcome measure and one analysis. It reports no posted secondary outcomes. This limits the scope of the outcome results available for interpretation, but does not establish whether other outcomes were collected or analyzed outside the posted record. Details of any planned multiplicity strategy, such as adjustment for multiple statistical tests, are not reported in the registry record. With one posted primary measure and analysis, the record does not describe a broader set of reported endpoint tests requiring interpretation together.

The primary analysis population is described as intention-to-treat with multiple imputation of missing outcomes. This indicates that missing outcomes were addressed through imputation in the intended analysis, but the registry does not report how many outcomes were missing, when they were missing, or whether missingness differed by group. The variables included in the imputation model, the number of imputed datasets, and the assumptions about missingness are not reported in the registry record.

The registry lists 62 participants in each outcome group. Those group counts alone do not disclose whether all participants had observed endpoint data or how imputation contributed to the reported analysis. No detailed accounting of randomized participants, exclusions, withdrawals, or follow-up completion is reported in the registry record. It is therefore not possible to assess the amount of missing information or to judge how sensitive the result might be to alternative missing-data assumptions.

The registry identifies the analysis as linear regression and non-inferiority, but does not report the covariates, model specification, or the non-inferiority margin. It also does not report interim analyses or the original statistical analysis plan. Those elements should not be inferred from the trial title or the posted p-value. The record supports a description of the analysis label and numerical result, but not a full independent audit of the analytic decisions.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: dizziness (1.6%), pneumonia (1.6%), hypertension (1.6%), fall (1.6%), limb fracture (1.6%).

For serious adverse events, the registry lists 1 affected participant among 62 at risk in the telerehabilitation group and 6 affected participants among 62 at risk in the in-clinic group. The posted deaths are 0 in each group. These are counts of participants reported as affected, not a comparison adjusted for exposure or follow-up time; such adjustment is not reported in the registry record.

The listed serious adverse event terms are Dizziness, Pneumonia, Hypertension, Fall, Limb fracture, and Subdural hemorrhage, each shown as 1.6%. The registry also reports other adverse events affecting 10 participants among 62 at risk in the telerehabilitation group and 7 among 62 at risk in the in-clinic group. The record does not provide enough detail in these figures to infer severity, timing, relatedness to treatment, or whether participants experienced more than one event.

These counts are descriptive safety information from the registry. They do not establish that the treatment setting caused or prevented an event, and they should not be read as a definitive comparative safety analysis. The registry record does not report event-level exposure time or a statistical comparison of adverse-event rates.

16. What the Result Does, and Does Not, Mean

The posted results describe average change in arm motor Fugl-Meyer score in participants assigned to home telerehabilitation and in-clinic therapy. The group means are 7.86 and 8.36, respectively. The registry’s regression analysis reports a mean difference (net) of 0.06, with a 95% confidence interval from -2.14 to 2.26. An unadjusted recalculation gives a mean difference of 0.5, with a 95% confidence interval from -1.916 to 2.916. Both intervals include zero and allow for differences in either direction.

The record labels the sponsor’s analysis as non-inferiority, but the non-inferiority margin is not reported in the registry record. Without that margin, readers cannot determine from the posted numerical result alone whether the estimate and interval satisfy the prespecified non-inferiority criterion. The high p-value does not establish equivalence, and failure to demonstrate a conventional difference would not by itself establish that two approaches are sufficiently similar.

The result also does not show that one setting causes a particular change for every participant. Group means summarize average outcomes and the standard deviations indicate variability. The confidence intervals describe uncertainty about group differences under the respective analyses; they do not describe the range of individual participant outcomes.

The trial record describes random assignment, a common planned therapy schedule, and different delivery settings. It also identifies an intention-to-treat analysis with multiple imputation. The posted information does not describe the missing-data pattern or imputation model in detail, nor does it fully specify the regression analysis. These reporting limits constrain independent assessment of the result.

Accordingly, the registry supports a careful account of the observed group summaries, the sponsor’s reported regression estimate, and the uncertainty around both that estimate and the unadjusted recalculation. It does not support a claim that one treatment works or does not work, or an independent conclusion about non-inferiority in the absence of the margin and decision rule.

17. Why This Trial Matters Statistically

This trial is statistically informative because it sets up a direct comparison of two delivery settings while describing a shared therapy schedule. The intervention and comparator were intended to provide the same intensity, duration, and frequency of arm motor therapy, with the setting and delivery system differing. That design makes the treatment-setting comparison the central statistical question rather than a comparison of different planned therapy schedules.

The endpoint is a continuous change score on a defined arm motor scale. Reporting group means and standard deviations makes the average changes and participant-level spread visible. The registry also posts a confidence interval and a p-value for its regression analysis, allowing readers to consider uncertainty rather than focusing only on point estimates.

The non-inferiority framing matters because it requires a different interpretation from a standard test for any difference. A reader needs the prespecified margin to understand what size of disadvantage would still be considered acceptable for the study objective. Since the margin is not reported in the registry record, the label alone is insufficient to independently verify the intended decision.

The separation between the sponsor’s regression result and the unadjusted recalculation also illustrates why analysis methods matter. A recalculation from group summaries can be useful as a transparent check, but it cannot reproduce a regression analysis without its model specification. The registry’s description of intention-to-treat with multiple imputation raises another important consideration: missing outcomes and the assumptions used to address them can affect uncertainty and interpretation, yet those details are not reported.

Finally, the posted safety counts offer descriptive context, while not constituting a complete comparative safety analysis. Taken together, the record illustrates the value of reporting effect estimates, intervals, analysis populations, and design details alongside p-values. It also shows why a stated trial objective needs its corresponding decision criteria to support a clear statistical conclusion.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several trial-specific details needed for a fuller appraisal are not reported in the registry record. The non-inferiority margin and its rationale are absent, so the primary decision criterion cannot be checked. The original sample-size justification is not reported, and the planning illustration based on the observed effect should not be mistaken for that justification.

The registry identifies the sponsor’s method as linear regression but does not report its covariates, model specification, or diagnostic checks. It also does not give the detailed procedure used for multiple imputation, the amount or pattern of missing outcome data, or a sensitivity analysis for missing-data assumptions. These omissions limit the ability to reproduce or stress-test the primary analysis.

The posted outcome groups each include 62 participants, but the record does not provide a detailed participant flow or clarify how those counts correspond to observed and imputed outcomes. The registry identifies the trial as randomized, single-masked, and assessor-blind in its official title, but details of sequence generation, allocation concealment, and masking procedures are not reported in the registry record.

The in-clinic arm description is incomplete in the record: it says that unsupervised home sessions were guided by an individual, but the description ends before identifying that guidance in full. The record also provides limited detail about adverse-event timing, treatment relatedness, and exposure. These gaps should be kept separate from the numerical results; they do not establish that a particular bias or safety problem occurred.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Telerehabilitation in the Home Versus Therapy In-Clinic for Patients With Stroke — NCT02360488.
• Registry record last updated 2020-02-28; status last verified 2020-02.
• Results first posted 2020-02-28.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

The registry describes a completed, randomized, parallel-group, single-masked trial of home telerehabilitation versus in-clinic therapy for people with stroke and arm motor deficits. The study planned 124 participants and lists 62 in each posted outcome group. Both groups were assigned 6 weeks of therapy, with 36 sessions planned over that period.

The primary endpoint was change in arm motor Fugl-Meyer score from baseline to the 30 Day Follow-Up Visit. The registry reports a non-inferiority linear regression result with a mean difference (net) of 0.06 and a 95% confidence interval from -2.14 to 2.26. An unadjusted recalculation from the group summaries gives a mean difference of 0.5 and a 95% confidence interval from -1.916 to 2.916. The non-inferiority margin and several analysis details are not reported in the registry record, limiting independent interpretation of the non-inferiority conclusion.